instruction-pilot-outputs-greedy

# Dataset Card for "instruction-pilot-outputs-greedy"

This dataset contains model outputs generated from the human demonstrations provided in [`HuggingFaceH4/instruction-pilot-prompts`](https://huggingface.co/datasets/HuggingFaceH4/instruction-pilot-prompts).

To convert each language model into a dialogue agent, we prepended the following [LangChain prompt](https://github.com/hwchase17/langchain/blob/bfabd1d5c0bf536fdd1e743e4db8341e7dfe82a9/langchain/chains/conversation/prompt.py#LL4C21-L9C7) to each input:

```
The following is a friendly conversation between a human and an AI. \
The AI is talkative and provides lots of specific details from its context. \
If the AI does not know the answer to a question, it truthfully says it does not know.

Human: {input}
AI:
```

For reproducibility purposes, we used deterministic text generation (`temperature=0`) and set `max_new_tokens=100` (which is about the mean lenght of the Self-Instruct outputs).

# “指令先导贪婪输出”数据集卡片

本数据集包含基于 [`HuggingFaceH4/instruction-pilot-prompts`](https://huggingface.co/datasets/HuggingFaceH4/instruction-pilot-prompts) 中提供的人类演示样本生成的模型输出。

为将每一个大语言模型（Large Language Model，LLM）转换为对话AI 智能体，我们在每条输入内容之前添加了如下 [LangChain 提示词（prompt）](https://github.com/hwchase17/langchain/blob/bfabd1d5c0bf536fdd1e743e4db8341e7dfe82a9/langchain/chains/conversation/prompt.py#LL4C21-L9C7)：

以下为人类与人工智能之间的友好对话。
该人工智能健谈且会基于自身上下文提供大量详实细节。
若人工智能无法回答某一问题，则会如实表明自身不知情。

人类：{input}
人工智能：

为确保可复现性，我们采用了确定性文本生成方式（`temperature=0`），并设置`max_new_tokens=100`（该参数值约等于自指令（Self-Instruct）输出的平均长度）。